This article explores different techniques to improve the impedance bandwidth of on-chip dipole antennas in the sub-THz frequency range. Increasing the area of the dipole antenna has shown considerable improvement in bandwidth. However, this violates the design rule checks (DRC) of the foundry. Various topologies, such as squared-slotted dipole, meandered-slotted dipole, and straight-slotted dipole antennas, are introduced and implemented to increase the width of the on-chip antennas and thus the impedance bandwidth while meeting the DRC rules. All three topologies show better performance in terms of providing improved bandwidth. The straight-slotted technique is adopted as it offers less complexity and flexibility. The behavior of the impedances for different widths implemented by the straight-slotted topology has been analyzed in detail. A 6-strip straight-slotted dipole antenna results in an ultra-wide impedance bandwidth ranging from 76–262 GHz with a fractional bandwidth of 110% and a gain of −0.6 dBi at 159 GHz, while occupying a small silicon area of . To the best of the authors’ knowledge, this is the highest fractional bandwidth that is reported to date at these frequencies.
This work presents an Exponential Variation based Particle Swarm Optimization (EV-PSO) algorithm to improve the convergence rate and find an optimal solution to analog circuit optimization problems in a constrained-driven environment. Existing evolutionary algorithms have a lower convergence rate leading to higher design time. This work introduces two novel parameters, and , into the velocity update equation. These parameters dynamically vary with the number of iterations. The algorithm was implemented on the Python platform. The results have shown that, in comparison to the considered existing methods, the exponential variation of the parameters and in the proposed algorithms have a larger rate of convergence. The proposed EV-PSO has a convergence rate of 27 iterations, which is 57.8%, 65.38%, and 59.1% better than the conventional PSO, differential evolution (DE) and genetic algorithm (GA) respectively. The typical design obtained from the optimal solution is verified through the simulation using 45-nm CMOS technology. The optimal solution presented in this work meets the desired input specifications within the specified constrained environment.